# Writing a "String#jindex" method to do the same for "index"

**URL:** <https://www.ruby-forum.com/t/writing-a-string-jindex-method-to-do-the-same-for-index/86765>\
**Category:** Ruby\
**Created:** [February 25, 2007, 6:21pm UTC](https://www.ruby-forum.com/t/writing-a-string-jindex-method-to-do-the-same-for-index/86765 "2007-02-25T18:21:24Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![greg](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/g/71c47a/32.png) [@greg](https://www.ruby-forum.com/u/greg)\
**Post date:** [February 25, 2007, 6:21pm UTC](https://www.ruby-forum.com/t/writing-a-string-jindex-method-to-do-the-same-for-index/86765/1 "2007-02-25T18:21:24Z")

</div>

So I’ve been using the ‘jcode’ library to get UTF-8 support and wanted  
to add a convenience method called “jindex” that would work with multi-  
byte characters in much the same way that “jlength” does.

The basic idea is to call the real “index” method to do the work, and  
then convert the returned byte-count index into a character-count  
index before returning it to the user. This works fine, but the  
problem is that I can’t find a way to propagate $~ back to the user.

That is, after calling this:

‘foo’.index /foo/

$~ is set and $~[0] is “foo”.

But when I do this.

‘foo’.jindex /foo/

$~ is nil. The following short irb session demonstrates what’s  
happening in the two cases:

irb(main):001:0\> class StringTest \< String  
irb(main):002:1\> def index foo  
irb(main):003:2\> r = super  
irb(main):004:2\> end  
irb(main):005:1\> end  
=\> nil  
irb(main):006:0\> StringTest.new(‘foo’).index /o/  
=\> 1  
irb(main):007:0\> $~[0]  
NoMethodError: undefined method `[]’ for nil:NilClass  
from (irb):7  
from :0  
irb(main):008:0\> String.new(‘foo’).index /o/  
=\> 1  
irb(main):009:0\> $~[0]  
=\> “o”

So I guess this is a more general question. If something inside a  
method sets $~, how do you propagate this back to the caller? There  
was a thread called “How to preserve $~ when extending Regexp#match”  
last year that touched on a similar issue ([http://groups.google.com/](http://groups.google.com/)  
group/comp.lang.ruby/browse\_thread/thread/73ccf213a9c0802d/  
a93554cd466acb44). This leads me to believe that the  
“special\_local\_set” function is being used under the hood by the  
“index” method and there will be no way for my “jindex” method to  
mimick its behaviour without also getting down into the C extension  
level, something I am not too keen to do as it is probably a bit over  
my head for now…

Are my worst fears right on this one?

Cheers,  
Greg

---

<div class="post-metadata">

**Author:** ![greg](https://www.ruby-forum.com/letter_avatar_proxy/v4/letter/g/71c47a/32.png) [@greg](https://www.ruby-forum.com/u/greg)\
**Post date:** [February 26, 2007, 11:11am UTC](https://www.ruby-forum.com/t/writing-a-string-jindex-method-to-do-the-same-for-index/86765/2 "2007-02-26T11:11:20Z")

</div>

On 25 feb, 18:16, “Greg H.” [removed\_email\_address@domain.invalid](mailto:removed_email_address@domain.invalid) wrote:

> level, something I am not too keen to do as it is probably a bit over  
> my head for now…

For the record, I decided to bite the bullet and try writing my first  
C extension for Ruby. It works and uses rb\_backref\_get and  
rb\_backref\_set to ensure that $~ gets correctly propagated back to the  
caller. If anyone knows of a clean, pure-Ruby way to do this (without  
a C extension) please let me know as it would be better from a  
portability/deployment perspective.

Cheers,  
Greg